Because the network is wholesale only (more than 50 retail providers sell on it), no independent assessment has ever measured what it actually changed on the ground. This analysis closes that gap. We matched anonymized Speedtest\u00ae results to the official map of the so-called \u201cIntervention Area\u201d to the official record of when service switched on across the rollout, observing the same rural communities before and after the fiber arrived.<\/p>\n
Six and a half years in, with over 490,000 premises passed<\/a>, equivalent to 87% of the current 566,000-premise rollout footprint, and construction under way or complete across the entire planned network, the answer is unambiguous: It worked, albeit at a high cost, faster and more completely than the program\u2019s defenders predicted, and the way it worked carries lessons for every government now chasing universal connectivity around the world.<\/p>\n Key Takeaways:<\/p>\n The state Intervention Area has seen one of the steepest sustained improvements in fixed performance we have measured in a developed market. Median download speed across the Intervention Area rose from 11.60 Mbps in 2019 to 156.99 Mbps over the year to mid-2026, a 13.5 times increase, while median upload rose from 4.67 Mbps to 49.48 Mbps. This reflects the generational leap from mostly copper-based DSL services to XGS-PON.<\/p>\n Ireland\u2019s rural-urban speed gap has nearly closed. In 2019, median urban download speeds were 4.8 times higher than those in the Intervention Area. By 2026, that difference had narrowed to 1.6 times. The Intervention Area now also records slightly faster median upload speeds than urban Ireland, likely reflecting the more symmetrical performance of fiber relative to Virgin Media\u2019s lingering cable and open eir\u2019s legacy VDSL networks in cities (both now being upgraded).<\/p>\n Performance transforms the month the fiber goes live, not gradually. Aligning our data to each locality\u2019s official switch-on date shows a flat pre-arrival baseline of roughly 27 Mbps for two full years, then an immediate jump past 150 Mbps within twelve months of activation, 5.7 times the pre-arrival level, with adoption visible in the data almost immediately after each area opens for orders.<\/p>\n The same connections, measured before and after, got roughly ten times faster. Tracking recurring testers within the same areas across their activation date shows the typical NBI-connected home\u2019s measured speed multiplied by 10.3, a like-for-like comparison that removes any compositional doubt about what changed.<\/p>\n Starlink has become the clearest real-time indicator of where the build has yet to reach. In townlands still waiting for fiber by mid-2026, Starlink accounts for 14.3% of fixed testing devices and is still climbing. In townlands served with fiber since 2021 and 2022, Starlink\u2019s share of samples has stayed pinned below 2% for five years. Simply put, when fiber arrives, demand for satellite appears to stop growing.<\/p>\n Two decades of failure made Ireland\u2019s rural premises the hardest problem in European broadband<\/p>\n Ireland\u2019s rural connectivity problem predates the NBP by a generation. The 1999 privatization of the incumbent operator (now eir) left rural access networks chronically underinvested through the broadband transition, and the State\u2019s first major intervention, the National Broadband Scheme of 2009 to 2014, contracted a mobile operator (Three) to deliver a minimum of just 2.3 Mbps across 1,028 rural electoral divisions at a cost of \u20ac223 million ($253 million)<\/a>. It was obsolete almost on arrival. The NBP, announced in 2012 with a projected subsidy of \u20ac175 million ($198 million) for universal 30 Mbps, spent seven years being redesigned upward as both the ambition and the true cost of serving the final premises became clear.<\/p>\n The underlying difficulty is settlement geography. Ireland has one of the most dispersed rural populations in Western Europe, dominated by one-off housing strung along country roads rather than clustered villages. The final Intervention Area contains under a quarter of the State\u2019s premises but covers 96% of its landmass, and reaching it requires more than 100,000 kilometers of fiber route, fewer than six premises for every kilometer built. No commercial business case survives that arithmetic, which is why two decades of market-led deployment, including an incumbent fiber program that reached 300,000 of the easier rural premises, stopped where the Intervention Area begins.<\/p>\n The procurement nearly collapsed under this same arithmetic. After the incumbent carved those 300,000 commercially viable premises out of the map in 2017, both remaining rival bidders withdrew, leaving a single consortium led by Granahan McCourt to negotiate the contract that was eventually signed in November 2019 (Spanish infrastructure investor Asterion acquired an 80% stake<\/a> in 2022). The structure that emerged is what makes Ireland internationally unique.<\/p>\n Australia\u2019s NBN built a state-owned network on a mix of technologies that traded performance for coverage and is now funding fiber upgrades<\/a>. The UK\u2019s subsidy-backed model spreads subsidies across many suppliers and technologies. The U.S. has run reverse auctions and state grant programs open to everything from cable to low Earth orbit satellite. Ireland alone contracted one wholesale-only operator to build full fiber to essentially every unserved premises, published the obligation on a public map, and left retail competition to ride on top.<\/p>\n The technology choices compounded the ambition. The network runs entirely on carrier-grade equipment from a single vendor, Nokia, using XGS-PON, a passive optical standard capable of 10 Gbps, rather than the cheaper GPON generation most European operators deployed first. The entry-level product was set at 500 Mbps from launch, a 2 Gbps tier followed in January 2022, and the State\u2019s own documentation describes the asset as future-proofed to 10 Gbps and beyond. Rural Ireland, in other words, was not given a catch-up network. It was given a newer access layer than most of urban Europe.<\/p>\n That ambition was bought at a price that remains the program\u2019s most contested feature. The overall subsidy is capped at \u20ac2.7 billion ($3.1 billion), comprising a \u20ac2.1 billion ($2.4 billion) base, a \u20ac500 million ($567 million) contingency and \u20ac100 million ($113 million) for VAT, according to the State\u2019s auditors<\/a>.<\/p>\n Official rollout statistics put the base subsidy at \u20ac3,569 ($4,047) to \u20ac3,946 ($4,475) per premises depending on county, with the cap implying a ceiling near \u20ac4,800 ($5,443) once contingency and VAT are included, figures far beyond the per-premises support typical of gap-funded schemes in denser European markets.<\/p>\n Academic work in Telecommunications Policy<\/a> attributes the escalation from the original \u20ac175 million ($198 million) projection to the collision between a full-fiber mandate and Ireland\u2019s low-density geography, compounded by a procurement that ended with a single bidder. Whether that premium was worth paying is precisely what our measured performance data can now answer.<\/p>\n A wholesale network with no name on it<\/p>\n The NBP\u2019s design choice that made it effective is the same one that made it hard to evaluate. NBI sells no retail service. Its fiber reaches the home, but the bill, the router, and the brand belong to one of the more than 50 retail providers that resell the network, from national brands to local operators that previously sold fixed wireless. Conventional provider-level statistics therefore cannot isolate the intervention at all.<\/p>\n Three properties of the rollout make this measurement possible here. First, the Intervention Area is precisely defined by the state\u2019s own public mapping, which makes the boundary for study unusually clean. Areas where commercial operators built were excluded by design, so high-speed fiber appearing inside the Intervention Area is, with limited exceptions, the state network. Second, NBI publishes an openly accessible service-availability record with dates, which provides the activation timing for every part of the rollout.<\/p>\n Third, the network\u2019s wholesale products carry characteristic download and upload pairings, an entry tier of 500 Mbps with a 50 Mbps upstream, a gigabit tier with 100 Mbps up, and a 2 Gbps tier with 200 Mbps up (with a 5 Gbps tier now emerging). Those upload ceilings leave unmistakable bands in our measured data, a fingerprint of the underlying infrastructure that survives the retail branding layered on top.<\/p>\n Our attribution holds up under checks. Connections matching the network\u2019s signature are essentially absent across the Intervention Area before each area\u2019s official activation date, appear within a month of it, and the geography of attributed connections tracks officially reported county-level connection counts very closely.<\/p>\n For governments planning subsidized networks, there is a quiet lesson here. Publishing granular, openly accessible coverage data, as Ireland does, is what makes independent verification of outcomes possible at all.<\/p>\n From 11.60 Mbps to 156.99 Mbps in the same townlands<\/p>\n In 2019, the Intervention Area was a connectivity desert by any modern standard. The median connection in the footprint measured 11.60 Mbps down and 4.67 Mbps up, 84% of homes measured below 30 Mbps, and effectively none reached 300 Mbps. These were the homes left behind by two decades of market-led deployment, reliant on long copper loops, fixed wireless links (like Imagine\u2019s earlier WiMAX service) , and in some cases nothing at all.<\/p>\n By the year to mid-2026 the same footprint posted a median of 156.99 Mbps down and 49.48 Mbps up. The share of homes below 30 Mbps has fallen from 84% to 14%, while 62% now measure above 100 Mbps and a quarter above 300 Mbps. The distribution tells the story more vividly than the median. The 2019 data forms a single low hump centered near 10 Mbps. By 2026 that hump has largely dissolved, replaced by mass concentrated between 100 and 500 Mbps with distinct peaks at the network\u2019s plan tiers, the statistical signature of a population moving from scarcity directly to abundance.<\/p>\n Two features distinguish this from ordinary market-driven improvement. The pre-arrival trend was flat, not improving, so the change cannot be attributed to background upgrades. And because the entry product on the network is 500 Mbps, rural homes are skipping the intermediate rungs that urban consumers climbed over a decade (through open eir\u2019s earlier VDSL\/FTTC and Virgin Media\u2019s HFC\/cable networks), jumping from single-digit megabits directly to gigabit-class service.<\/p>\n The nature of the National Broadband Plan contract means that NBI sets one wholesale price for each mandated product\/speed tier, capped against the equivalent wholesale product in the commercially served \u201cExcluded Area\u201d (i.e. open eir services in urban Ireland).<\/p>\n For example, NBI\u2019s published consumer wholesale tariff is currently \u20ac29.49 ($33.44) per month excluding taxes for 500\/50 Mbps and \u20ac34.49 ($39.11) for 1 Gbps\/100 Mbps, before interconnect and traffic charges. This pricing model was designed to help retailers offer rural consumers broadly comparable propositions to urban customers and support a level urban-rural playing field.<\/p>\n Speeds jump the month the fiber goes live<\/p>\n Aligning each locality to its official switch-on date produces the cleanest picture of the intervention\u2019s mechanics. For two years before activation, measured medians sit flat at roughly 27 Mbps. In the month service opens, performance inflects vertically. Within a year, the median sits above 150 Mbps, and it keeps rising beyond that as households migrate up the tier ladder toward gigabit plans.<\/p>\n The like-for-like view is just as stark. Following recurring testers in the same areas across activation, the typical connected home saw its measured speed multiply by 10.3. Households that previously endured single-digit DSL or congested fixed wireless now sit on the same access technology as the best-served urban premises in the country, and in many cases on a newer one.<\/p>\n NBI reports<\/a> average take-up of around 37% of passed premises, exceeding 60% in the areas live longest. The speed with which the performance signature appears after each area opens suggests pent-up demand rather than slow discovery. Rural Ireland was not indifferent to broadband (it was waiting for it).<\/p>\n The upload crossover puts rural homes ahead of urban Ireland<\/p>\n The most consequential finding in our analysis for the long run is on the uplink side. NBI\u2019s fiber tiers carry 50, 100, and 200 Mbps upstream, and the growing share of rural households on gigabit-class plans pulls the rural distribution upward. Median upload speeds across the Intervention Area overtook urban Ireland during 2026, at around 53 Mbps vs. around 51 Mbps, and the gap is widening.<\/p>\n While the prevalence is declining as the fiber rollout advances at pace, as mentioned, parts of urban Ireland\u2019s upload medians still lean on Virgin Media\u2019s cable infrastructure (and open eir\u2019s VDSL), whose asymmetric design caps most uplink plans near 50 Mbps.<\/p>\n Upload capacity is the dimension that matters increasingly for remote work, cloud backup, agricultural telemetry (important in rural Ireland), video calling, and small-business hosting, including emerging voice and video AI models. A state intervention designed for the hardest-to-serve premises has, on this dimension, leapfrogged the commercial networks serving Irish cities. For a program justified largely on regional development grounds, this inversion is the strongest evidence available that the investment bought durable infrastructure rather than catch-up.<\/p>\n The bottleneck is shifting from the doorstep to the router<\/p>\n As more rural households migrate from NBI\u2019s entry 500\/50 Mbps tier toward its gigabit and 2 Gbps plans, the router is becoming the constraint on what that connection can actually deliver inside the home. A Wi-Fi 5 device running on a single 5 GHz access point, still the dominant configuration across most NBI retailers\u2019 installed bases, caps the wireless link well short of what the fiber backhaul is capable of carrying.<\/p>\n Wi-Fi 6, 6E and 7 offer the spectral efficiency and multi-user performance gains required to realize the full capacity of a gigabit plan, and that gap widens further on the 2 Gbps tier where Wi-Fi 5 is structurally inadequate. In practice, two households on identical NBI plans in the same townland may therefore experience materially different speeds depending on which retailer supplied their router and when.<\/p>\n Our measured data shows the pace of Wi-Fi 6+ deployment varies sharply across the four largest NBI retailers by sample volume. Digiweb moved fastest: its measured base was already 70% Wi-Fi 6 or newer by Q2 2026, having crossed the 50% mark in mid-2024 (roughly a year ahead of its larger rivals).<\/p>\n Eir, meanwhile, crossed the 50% threshold in Q2 2026, while Vodafone reached 40%, having been essentially flat at around 11% through most of 2022 to 2024 before accelerating sharply from late 2025. Sky is the clearest laggard, still 73% Wi-Fi 5 at Q2 2026 despite being one of the largest NBI retailers by connection count. All four are riding the same XGS-PON infrastructure; the divergence in Wi-Fi generation reflects router procurement decisions and CPE refresh cycles.<\/p>\n It should be noted that performance can also differ for reasons beyond in-home Wi-Fi. While NBI supplies the common wholesale access network, each retailer determines the customer\u2019s IP\/core-network integration, transit and peering. Sky, for example, historically routed some traffic through the UK, resulting in markedly poorer application latency for its customers and contributing to IP geolocation problems<\/a> for content access.<\/p>\n Starlink thrives only where the fiber has not yet arrived<\/p>\n Low Earth orbit (LEO) satellite is often framed as an alternative to rural fiber programs. Ireland now offers a controlled comparison of the two, because the staggered rollout created neighboring populations with and without fiber access at every point in time. The pattern is unambiguous. In townlands still unserved by mid-2026, LEO provider Starlink has grown relentlessly and now accounts for 14.3% of fixed samples, approaching one in seven. In townlands where fiber arrived in 2021 and 2022, Starlink\u2019s share has remained below 2% for five years, with no growth despite the same marketing, pricing, and availability.<\/p>\n Satellite subscriptions nationally more than doubled in the year to late 2025 on the regulator\u2019s figures<\/a> (with most of the growth accruing to Starlink), which is sometimes read as a verdict on the fiber program. The footprint-level view shows the opposite. The growth is concentrated almost entirely in the shrinking territory the build has not yet reached, and it stops where the fiber starts (although noting LEO is increasingly being used in urban and suburban areas as a resilience back up and for other use cases requiring on-demand connectivity or mobility or similar). Together, the Irish example shows LEO functioning as a bridge technology and a real-time map of unmet demand, not as a substitute for wireline investment.<\/p>\n What other governments can take from the Irish model, including its costs<\/p>\n The Irish design concentrated execution risk in one counterparty and one technology, and its early delays and cost critiques were real. But it produced something rare, a rural access layer that is uniform, multi-gigabit-capable, future-proofed, and competitively neutral, with dozens of retailers competing on identical infrastructure from day one. Our measured outcomes above are the dividend on that design. The honest ledger for other governments has two further entries, what it cost, and who it displaced.<\/p>\n On cost, as mentioned earlier, the per-premises arithmetic will look daunting to any treasury. A base subsidy of roughly \u20ac3,700 ($4,196) to \u20ac3,950 ($4,479) per premises, and a ceiling near \u20ac4,800 ($5,443) with contingency and VAT, is multiples of what gap-funded programs in denser geographies pay.<\/p>\n The counterargument is, however, visible in our data. Cheaper mixed-technology interventions bought slower networks that are already being upgraded at additional cost, while Ireland\u2019s premium bought an asset with 10 Gbps headroom and a measured performance dividend that arrived the month each area went live. A relevant comparison is not just the cheapest subsidy per premises but the cost per decade of adequacy (and the sovereignty of the underlying infrastructure), and on that measure the Irish premium looks materially better than it did in 2019.<\/p>\n The displacement effects deserve equal attention, because the Intervention Area was never empty. Dozens of regional fixed wireless operators had spent two decades building businesses in exactly the territory the State elected to overbuild with subsidized fiber, and the largest of them (Imagine), valued at around \u20ac200 million<\/a> ($227 million) when international capital bought in during 2018 and planning a \u20ac300 million ($340 million)<\/a> next-generation wireless network, objected that it was excluded from the tender and that the full-fiber mandate was designed around it.<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c0_hero_explainer.gif\")
<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c7_map_pair-1-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c5_fingerprint-1-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c2_dist_morph-1.gif\")
<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c4_eventstudy-1-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c3_convergence-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/nbi_wifi_generation_chart-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c6_starlink-1-scaled.png\")
<\/a><\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/ie\/wp-content\/uploads\/2026\/07\/c1_hero_map-1.gif\")
<\/p>\n